We have featured thermal camera projects by [Max Ritter] before, but [Max] has just taken the next step: he is offering the latest version as a build-it-yourself kit. The DIY Thermocam improves on his previous designs by capturing 60 by 80 pixel thermal images, which can be combined with visible light images from an accompanying 640 by 480 pixel camera to produce the final image. It is built around the FLIR Lepton module that has been used in many of the recent commercial thermal cameras that we have seen. Max has also added a battery and display, making the whole thing a standalone camera.
The firmware that runs all this is open-source and written in C++ for easy modification, so users can build their own thermal camera.”The approach is to offer people the self-assembly kit so that they can use it as a development platform to do whatever they want to achieve with thermal imaging”[Max] told us. The kit runs €429 (about $468), with free shipping worldwide.
44 thoughts on “Build Your Own Thermal Camera”
I see an “Open Source” logo on the site but no source files for hardware or firmware. Are they available somewhere?
You could search for “DIY-Thermocam” on github.com or which gives one newly created empty repository. Which I would guess they are holding off populating until they have shipped some units.
I would guess that most of the information is currently spread over all their downloads http://www.diy-thermocam.net/downloads/ just because it is open, does not always mean that it will be Zero effort to replicate.
You could also search for “FLIR Lepton Arduino” on github.com and I’m sure you will find some useful info.
Just as an FYI the Open Source logo =! Open Hardware logo
Exactly :) The software is open-source and on github in the meantime (https://github.com/maxritter/DIY-Thermocam), the hardware is not.
It’s a shame that this technology is not much, much cheaper. It has quite a lot of usages, I could list 20-30 almost off the bat. 400$? Meh. Not available to a Eastern Europe robotics student. I wish thermal imaging was cheaper, but as for now, it’s >100$. https://hackaday.io/project/1974-tj-99-thermal-imager – I’m still watching on this.
A 16×4 pixels sensor will be much cheaper than 60×80 pixel sensor. What could be very interesting would be adding the stepper motor from that project to get 60×320 pixel image with the FLIR sensor.
A pan-tilt servo and voila! Thermo-radar or thermidar!
The FLIR One for iPhone is around $110 or so on sale. You do need an iPhone or iPad with Lightning connector to use it, of course. (A simple mechanical mod is needed to use it with iPad.) You can also take the unit apart and use the sensor or perhaps reverse engineer the interface to connect to something else.
Just bought one from eBay for my 5S phone for $160.
FLIR has at least one patent on combining images like this so they can probably expect a C&D really quick.
Patent, schmatent, how the hell can stitching images together warrant a patent, the world has gone mad. I’ve been combining digital images since the days of the 4116 Dynamic ram as image sensor. http://hackaday.com/2014/04/05/taking-pictures-with-a-dram-chip/ Sticking pictures side by side, even thermal pictures… and patenting the idea… give me strength.
Did you combine it with a thermal imager? Then you should have patented it.
Like it or not, they did.
DIY is not subject to patent law. They aren’t violating any patents by selling you a bunch of parts and you using some open source software downloaded on the internet to make it work.
Hahaha. Oh yes it is.
No, they are not. Like Garbz said, you can do it for yourself, you just can’t sell a patent infridging machine, so they can sell the parts, you can put a software in there that “infridges” the patent, but as long as you don’t sell it like that, you and everyone else is ok.
J and Garbz are correct. Selling a kit of parts does not infringe on a design patent. The software however, would need to be given away for free, or sold as non-functional source code that requires the DIY’er to fix it to make it work. The open source software makes no difference here. It does not matter if it was written from scratch with no knowledge of the FLIR software or patent. But software patents are very difficult to enforce, so the guy would probably be fine with selling it, especially if the FLIR patent describes the particular operating system or embedded processor the software is running on.
WANT. But without the framerate cap- for realtime thermal video. ITAR has good reason to restrict that though. Sigh.
ITAR would make sense if only NATO members wouldn’t be exporting dual-use technologies to countries like Russia. Honestly, in many areas such restrictions are pain in the butt and should become a thing of the past. Did you know for instance that modern high-end CNC machines have a GPS position lock and will stop working if you move them even one meter?
We bought a new Makino Mag 1 at one of the shop I worked at and it had this built in. If you move it Makino has to come out to reset it. Not that you would ever want to move that 60,000lb machine unless you really had to, it took about a month after setting it on the floor to get it all hooked up and running.
For what pvrpose?
Basically high end machine tools were sold to embargoed countries (first russia in the 80s, where they were claimed to be used to manufacture the drive of the Akula-class submarines, and more recently Iran). The gps/gyro lockout (which requires the manufacturer to confirm your location if the machine is moved at all) is supposed to verify that their machine didn’t end up in Iran or North Korea.
I fully support the notion of open source and kits, but why is this that expensive? FLIR’s own model TG165 is a robust unit with the similar or better specs than this kit, comes ready-to-use out of the box, with SD card storage, battery, and charger. For $400. Is FLIR gouging sensor buyers to keep their own market?
indeed, the TG165 also features 80×60 pixels and costs €300 on RS.
This is a nice open source project, but way too expensive.
I agree, but these guys aren’t making much on the project — the list price of the FLiR Lepton is $175 t any quantity. FLiR is the one making the money.
FLIR ONE does 160×120 and is €270
So it costs as much as a real entry level flir product.. why bother?
The disappointing part is that the video and combined pics must be made manually in post-processing. The description makes it sound like it is real time. Also, I see nothing about frame rate. The sensor is capable of 8.6 Hz, but how fast does this camera work?
Okay, here’s the dirty secret – the FLIR C2 (~$US 700) will output 30 fps as a webcam. Save yourselves the trouble. We went down this road and it’s a mess.
A Seek Thermal offer an higher resolution at about half the price. This project is still to expensive to be competitive.
Hey guys, I am Max Ritter, the creator of the DIY-Thermocam.
Thanks for your feedback! I am just a 23 old student that does all of this without the help of any company or external partner at my little home. So of course, I can not compete with big companies like FLIR or Seek Thermal in terms of pricing as I have to buy all of the single components on my own for normal market prices. The project is the result of my bachelor thesis, and for this, I think it’s a good result.
The github repo was not ready when this article has been posted, now you can see the firmware development here: https://github.com/maxritter/DIY-Thermocam
I would be very happy if some people would support my student project by buying one of the kits.
What always strikes me as weird is how seldom they use thermal imaging for medical diagnostics, both in human and veterinarian application. And with veterinarian you already have the issue that you can’t communicate with the patient, so it might be rather useful if you could get some objective information.
Many have tried. There is substantial research literature on it, and it’s a perennial grant application topic. Several companies have tried and failed to get a product to market. The biggest draw has been cancer diagnosis, in particular breast cancer. The idea is that active cancer, with its high metabolic rate, is warmer than normal tissue.
Unfortunately it just doesn’t work. It’s not reliable or a robust indication of anything. It is confounded by clothing, local environment (drafts, radiant sources and sinks) and recent history (e.g. how long since the patient removed their bra). Thermal IR can see only 50 microns into the body, or just a few cell layers into the top layer of skin. It gives no useful information of anything deeper.
Well it might not work for cancer, but why aim so high? I think it can certainly be useful to see injuries to ligatures and such (for sports diagnosis amongst others), and maybe if you make a endoscope type of deal some ailments to the throat like the vocal chord. Which might be worthwhile just by the amount of money going around in the music industry and how elusive it apparently can be when singers suddenly develop some voice issue.
And it doesn’t have to look deep into tissue, because the body often reacts with inflammation which should be visible on the outside, enough at least to tell if there is local inflammation and perhaps to center on the source, say for instance with back injuries.
I’m not a doctor but I think it must have a decent use, as long as you don’t expect it to be some SciFi miracle scanner.
And of course once you have the camera a scan is basically free, unlike many other scans which always cost a lot of money and/or expose the patient to radiation.
A human hair is about 10 microns in diameter, 50 microns is not really all that far to be able to see into a body. If there is damage within the first 50 microns it will be detectable by touch.
correction: 17 microns is a human hairs diameter.
17 um is awfully fine for human hair. Maybe on some areas. Head hair is more like 50 um. I can’t speak for the rest of humanity or for other body parts, but everyone in my family has head hair diameters between 50 and 55 um. (Yes, we measured. With frickin’ lasers. I’m a physicist, and that’s what we do for fun around our house.)
@Paul I got my number from here – https://en.wikipedia.org/wiki/10_micrometres “17 µm – minimum width of a strand of human hair” which is useful but probably not as a much fun as taking your own actual measurements.
But if an area of your body heats up the heats will be radiated out, it will in fact be beyond the damn surface of the skin and not suddenly stop.
@Whatnot: Radiation has no significant contribution to heat distribution in the body: it cannot be “radiated out”, as it only propagates 50 um or so, as said earlier. Even conduction is not that significant: heat transport in the body is dominated by blood flow. Blood vessels in the skin are a separate system (and separately controlled) from the those in the underlying organs or muscles, so skin temperature measurements are a poor indicator of the health of underlying tissue, which is why all those medical thermal imaging product proposals die before their first clinical trial. Doesn’t stop the vulture capitalists from being suckered into trying though…
Paul, you should make a project of how to measure a human hair with lasers. That would be pretty cool
@Jehu: no need for me to do it. There are lots of examples on the web now, such as student.societyforscience.org/blog/eureka-lab/measure-width-your-hair-laser-pointer
Hey Max, have you tried to put a ZnSe lens on the sensor to increase the range? Thanks!
Yes, that should work :)
I once used a cheap ZnSe CO2 lense from Ebay to get better focus for close-work applications like PCB inspection.
Please be kind and respectful to help make the comments section excellent. (Comment Policy)